2 ******************************************************************************
4 * Copyright (C) 2000-2010, International Business Machines
5 * Corporation and others. All Rights Reserved.
7 ******************************************************************************
10 * tab size: 8 (not used)
13 * created on: 2000jun29
14 * created by: Markus W. Scherer
16 * Arabic letter shaping implemented by Ayman Roshdy
19 #include "unicode/utypes.h"
20 #include "unicode/uchar.h"
21 #include "unicode/ustring.h"
22 #include "unicode/ushape.h"
26 #include "ubidi_props.h"
30 * This implementation assumes that the internal encoding is UTF-16
31 * or UTF-32, not UTF-8.
32 * The main assumption is that the Arabic characters and their
33 * presentation forms each fit into a single UChar.
34 * With UTF-8, they occupy 2 or 3 bytes, and more than the ASCII
37 # error This implementation assumes UTF-16 or UTF-32 (check UTF_SIZE)
41 * ### TODO in general for letter shaping:
42 * - the letter shaping code is UTF-16-unaware; needs update
43 * + especially invertBuffer()?!
44 * - needs to handle the "Arabic Tail" that is used in some legacy codepages
45 * as a glyph fragment of wide-glyph letters
46 * + IBM Unicode conversion tables map it to U+200B (ZWSP)
47 * + IBM Egypt has proposed to encode the tail in Unicode among Arabic Presentation Forms
50 /* definitions for Arabic letter shaping ------------------------------------ */
60 #define COMBINE (SHADDA+CSHADDA)
62 #define HAMZAFE_CHAR 0xfe80
63 #define HAMZA06_CHAR 0x0621
64 #define YEH_HAMZA_CHAR 0x0626
65 #define YEH_HAMZAFE_CHAR 0xFE89
66 #define LAMALEF_SPACE_SUB 0xFFFF
67 #define TASHKEEL_SPACE_SUB 0xFFFE
68 #define NEW_TAIL_CHAR 0xFE73
69 #define OLD_TAIL_CHAR 0x200B
70 #define LAM_CHAR 0x0644
71 #define SPACE_CHAR 0x0020
72 #define SHADDA_CHAR 0xFE7C
73 #define TATWEEL_CHAR 0x0640
74 #define SHADDA_TATWEEL_CHAR 0xFE7D
77 #define DESHAPE_MODE 1
79 static UChar tailChar = OLD_TAIL_CHAR;
80 static uint32_t uShapeLamalefBegin = U_SHAPE_LAMALEF_BEGIN;
81 static uint32_t uShapeLamalefEnd = U_SHAPE_LAMALEF_END;
82 static uint32_t uShapeTashkeelBegin = U_SHAPE_TASHKEEL_BEGIN;
83 static uint32_t uShapeTashkeelEnd = U_SHAPE_TASHKEEL_END;
84 static int spacesRelativeToTextBeginEnd = 0;
86 static const uint8_t tailFamilyIsolatedFinal[] = {
103 static const uint8_t tashkeelMedial[] = {
122 static const UChar yehHamzaToYeh[] =
124 /* isolated*/ 0xFEEF,
128 static const uint8_t IrrelevantPos[] = {
134 static const UChar convertLamAlef[] =
146 static const UChar araLink[178]=
148 1 + 32 + 256 * 0x11,/*0x0622*/
149 1 + 32 + 256 * 0x13,/*0x0623*/
150 1 + 256 * 0x15,/*0x0624*/
151 1 + 32 + 256 * 0x17,/*0x0625*/
152 1 + 2 + 256 * 0x19,/*0x0626*/
153 1 + 32 + 256 * 0x1D,/*0x0627*/
154 1 + 2 + 256 * 0x1F,/*0x0628*/
155 1 + 256 * 0x23,/*0x0629*/
156 1 + 2 + 256 * 0x25,/*0x062A*/
157 1 + 2 + 256 * 0x29,/*0x062B*/
158 1 + 2 + 256 * 0x2D,/*0x062C*/
159 1 + 2 + 256 * 0x31,/*0x062D*/
160 1 + 2 + 256 * 0x35,/*0x062E*/
161 1 + 256 * 0x39,/*0x062F*/
162 1 + 256 * 0x3B,/*0x0630*/
163 1 + 256 * 0x3D,/*0x0631*/
164 1 + 256 * 0x3F,/*0x0632*/
165 1 + 2 + 256 * 0x41,/*0x0633*/
166 1 + 2 + 256 * 0x45,/*0x0634*/
167 1 + 2 + 256 * 0x49,/*0x0635*/
168 1 + 2 + 256 * 0x4D,/*0x0636*/
169 1 + 2 + 256 * 0x51,/*0x0637*/
170 1 + 2 + 256 * 0x55,/*0x0638*/
171 1 + 2 + 256 * 0x59,/*0x0639*/
172 1 + 2 + 256 * 0x5D,/*0x063A*/
173 0, 0, 0, 0, 0, /*0x063B-0x063F*/
175 1 + 2 + 256 * 0x61,/*0x0641*/
176 1 + 2 + 256 * 0x65,/*0x0642*/
177 1 + 2 + 256 * 0x69,/*0x0643*/
178 1 + 2 + 16 + 256 * 0x6D,/*0x0644*/
179 1 + 2 + 256 * 0x71,/*0x0645*/
180 1 + 2 + 256 * 0x75,/*0x0646*/
181 1 + 2 + 256 * 0x79,/*0x0647*/
182 1 + 256 * 0x7D,/*0x0648*/
183 1 + 256 * 0x7F,/*0x0649*/
184 1 + 2 + 256 * 0x81,/*0x064A*/
185 4 + 256 * 1, /*0x064B*/
186 4 + 128 + 256 * 1, /*0x064C*/
187 4 + 128 + 256 * 1, /*0x064D*/
188 4 + 128 + 256 * 1, /*0x064E*/
189 4 + 128 + 256 * 1, /*0x064F*/
190 4 + 128 + 256 * 1, /*0x0650*/
191 4 + 64 + 256 * 3, /*0x0651*/
192 4 + 256 * 1, /*0x0652*/
193 4 + 256 * 7, /*0x0653*/
194 4 + 256 * 8, /*0x0654*/
195 4 + 256 * 8, /*0x0655*/
196 4 + 256 * 1, /*0x0656*/
197 0, 0, 0, 0, 0, /*0x0657-0x065B*/
198 1 + 256 * 0x85,/*0x065C*/
199 1 + 256 * 0x87,/*0x065D*/
200 1 + 256 * 0x89,/*0x065E*/
201 1 + 256 * 0x8B,/*0x065F*/
202 0, 0, 0, 0, 0, /*0x0660-0x0664*/
203 0, 0, 0, 0, 0, /*0x0665-0x0669*/
204 0, 0, 0, 0, 0, 0, /*0x066A-0x066F*/
205 4 + 256 * 6, /*0x0670*/
206 1 + 8 + 256 * 0x00,/*0x0671*/
211 1, 1, /*0x0676-0x0677*/
212 1+2, 1+2, 1+2, 1+2, 1+2, 1+2, /*0x0678-0x067D*/
213 1+2+8+256 * 0x06, 1+2, 1+2, 1+2, 1+2, 1+2, /*0x067E-0x0683*/
214 1+2, 1+2, 1+2+8+256 * 0x2A, 1+2, /*0x0684-0x0687*/
215 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /*0x0688-0x0691*/
216 1, 1, 1, 1, 1, 1, 1+8+256 * 0x3A, 1, /*0x0692-0x0699*/
217 1+2, 1+2, 1+2, 1+2, 1+2, 1+2, /*0x069A-0x06A3*/
218 1+2, 1+2, 1+2, 1+2, /*0x069A-0x06A3*/
219 1+2, 1+2, 1+2, 1+2, 1+2, 1+2+8+256 * 0x3E, /*0x06A4-0x06AD*/
220 1+2, 1+2, 1+2, 1+2, /*0x06A4-0x06AD*/
221 1+2, 1+2+8+256 * 0x42, 1+2, 1+2, 1+2, 1+2, /*0x06AE-0x06B7*/
222 1+2, 1+2, 1+2, 1+2, /*0x06AE-0x06B7*/
223 1+2, 1+2, 1+2, 1+2, 1+2, 1+2, /*0x06B8-0x06BF*/
224 1+2, 1+2, /*0x06B8-0x06BF*/
227 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /*0x06C2-0x06CB*/
228 1+2+8+256 * 0xAC, /*0x06CC*/
230 1+2, 1+2, 1+2, 1+2, /*0x06CE-0x06D1*/
231 1, 1 /*0x06D2-0x06D3*/
234 static const uint8_t presALink[] = {
235 /***********0*****1*****2*****3*****4*****5*****6*****7*****8*****9*****A*****B*****C*****D*****E*****F*/
236 /*FB5*/ 0, 1, 0, 0, 0, 0, 0, 1, 2,1 + 2, 0, 0, 0, 0, 0, 0,
237 /*FB6*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
238 /*FB7*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 2,1 + 2, 0, 0,
239 /*FB8*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1,
240 /*FB9*/ 2,1 + 2, 0, 1, 2,1 + 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
241 /*FBA*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
242 /*FBB*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
243 /*FBC*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
244 /*FBD*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
245 /*FBE*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
246 /*FBF*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 2,1 + 2,
247 /*FC0*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
248 /*FC1*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
249 /*FC2*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
250 /*FC3*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
251 /*FC4*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
252 /*FC5*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 4, 4,
256 static const uint8_t presBLink[]=
258 /***********0*****1*****2*****3*****4*****5*****6*****7*****8*****9*****A*****B*****C*****D*****E*****F*/
259 /*FE7*/1 + 2,1 + 2,1 + 2, 0,1 + 2, 0,1 + 2,1 + 2,1 + 2,1 + 2,1 + 2,1 + 2,1 + 2,1 + 2,1 + 2,1 + 2,
260 /*FE8*/ 0, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 2,1 + 2, 0, 1, 0,
261 /*FE9*/ 1, 2,1 + 2, 0, 1, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,
262 /*FEA*/1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 0, 1, 0, 1, 0,
263 /*FEB*/ 1, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,
264 /*FEC*/1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,
265 /*FED*/1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,
266 /*FEE*/1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 2,1 + 2, 0, 1, 0,
267 /*FEF*/ 1, 0, 1, 2,1 + 2, 0, 1, 0, 1, 0, 1, 0, 1, 0, 0, 0
270 static const UChar convertFBto06[] =
272 /***********0******1******2******3******4******5******6******7******8******9******A******B******C******D******E******F***/
273 /*FB5*/ 0x671, 0x671, 0, 0, 0, 0, 0x07E, 0x07E, 0x07E, 0x07E, 0, 0, 0, 0, 0, 0,
274 /*FB6*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
275 /*FB7*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0x686, 0x686, 0x686, 0x686, 0, 0,
276 /*FB8*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0x698, 0x698, 0, 0, 0x6A9, 0x6A9,
277 /*FB9*/ 0x6A9, 0x6A9, 0x6AF, 0x6AF, 0x6AF, 0x6AF, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
278 /*FBA*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
279 /*FBB*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
280 /*FBC*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
281 /*FBD*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
282 /*FBE*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
283 /*FBF*/ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0x6CC, 0x6CC, 0x6CC, 0x6CC
286 static const UChar convertFEto06[] =
288 /***********0******1******2******3******4******5******6******7******8******9******A******B******C******D******E******F***/
289 /*FE7*/ 0x64B, 0x64B, 0x64C, 0x64C, 0x64D, 0x64D, 0x64E, 0x64E, 0x64F, 0x64F, 0x650, 0x650, 0x651, 0x651, 0x652, 0x652,
290 /*FE8*/ 0x621, 0x622, 0x622, 0x623, 0x623, 0x624, 0x624, 0x625, 0x625, 0x626, 0x626, 0x626, 0x626, 0x627, 0x627, 0x628,
291 /*FE9*/ 0x628, 0x628, 0x628, 0x629, 0x629, 0x62A, 0x62A, 0x62A, 0x62A, 0x62B, 0x62B, 0x62B, 0x62B, 0x62C, 0x62C, 0x62C,
292 /*FEA*/ 0x62C, 0x62D, 0x62D, 0x62D, 0x62D, 0x62E, 0x62E, 0x62E, 0x62E, 0x62F, 0x62F, 0x630, 0x630, 0x631, 0x631, 0x632,
293 /*FEB*/ 0x632, 0x633, 0x633, 0x633, 0x633, 0x634, 0x634, 0x634, 0x634, 0x635, 0x635, 0x635, 0x635, 0x636, 0x636, 0x636,
294 /*FEC*/ 0x636, 0x637, 0x637, 0x637, 0x637, 0x638, 0x638, 0x638, 0x638, 0x639, 0x639, 0x639, 0x639, 0x63A, 0x63A, 0x63A,
295 /*FED*/ 0x63A, 0x641, 0x641, 0x641, 0x641, 0x642, 0x642, 0x642, 0x642, 0x643, 0x643, 0x643, 0x643, 0x644, 0x644, 0x644,
296 /*FEE*/ 0x644, 0x645, 0x645, 0x645, 0x645, 0x646, 0x646, 0x646, 0x646, 0x647, 0x647, 0x647, 0x647, 0x648, 0x648, 0x649,
297 /*FEF*/ 0x649, 0x64A, 0x64A, 0x64A, 0x64A, 0x65C, 0x65C, 0x65D, 0x65D, 0x65E, 0x65E, 0x65F, 0x65F
300 static const uint8_t shapeTable[4][4][4]=
302 { {0,0,0,0}, {0,0,0,0}, {0,1,0,3}, {0,1,0,1} },
303 { {0,0,2,2}, {0,0,1,2}, {0,1,1,2}, {0,1,1,3} },
304 { {0,0,0,0}, {0,0,0,0}, {0,1,0,3}, {0,1,0,3} },
305 { {0,0,1,2}, {0,0,1,2}, {0,1,1,2}, {0,1,1,3} }
309 * This function shapes European digits to Arabic-Indic digits
310 * in-place, writing over the input characters.
311 * Since we know that we are only looking for BMP code points,
312 * we can safely just work with code units (again, at least UTF-16).
315 _shapeToArabicDigitsWithContext(UChar *s, int32_t length,
317 UBool isLogical, UBool lastStrongWasAL) {
318 const UBiDiProps *bdp;
322 bdp=ubidi_getSingleton();
325 /* the iteration direction depends on the type of input */
327 for(i=0; i<length; ++i) {
329 switch(ubidi_getClass(bdp, c)) {
330 case U_LEFT_TO_RIGHT: /* L */
331 case U_RIGHT_TO_LEFT: /* R */
332 lastStrongWasAL=FALSE;
334 case U_RIGHT_TO_LEFT_ARABIC: /* AL */
335 lastStrongWasAL=TRUE;
337 case U_EUROPEAN_NUMBER: /* EN */
338 if(lastStrongWasAL && (uint32_t)(c-0x30)<10) {
339 s[i]=(UChar)(digitBase+c); /* digitBase+(c-0x30) - digitBase was modified above */
347 for(i=length; i>0; /* pre-decrement in the body */) {
349 switch(ubidi_getClass(bdp, c)) {
350 case U_LEFT_TO_RIGHT: /* L */
351 case U_RIGHT_TO_LEFT: /* R */
352 lastStrongWasAL=FALSE;
354 case U_RIGHT_TO_LEFT_ARABIC: /* AL */
355 lastStrongWasAL=TRUE;
357 case U_EUROPEAN_NUMBER: /* EN */
358 if(lastStrongWasAL && (uint32_t)(c-0x30)<10) {
359 s[i]=(UChar)(digitBase+c); /* digitBase+(c-0x30) - digitBase was modified above */
371 *Function : This function inverts the buffer, it's used
372 * in case the user specifies the buffer to be
373 * U_SHAPE_TEXT_DIRECTION_LOGICAL
376 invertBuffer(UChar *buffer,int32_t size,uint32_t options,int32_t lowlimit,int32_t highlimit) {
379 for(i=lowlimit,j=size-highlimit-1;i<j;i++,j--) {
381 buffer[i] = buffer[j];
387 *Name : changeLamAlef
388 *Function : Converts the Alef characters into an equivalent
389 * LamAlef location in the 0x06xx Range, this is an
390 * intermediate stage in the operation of the program
391 * later it'll be converted into the 0xFExx LamAlefs
392 * in the shaping function.
394 static U_INLINE UChar
395 changeLamAlef(UChar ch) {
411 *Function : Resolves the link between the characters as
412 * Arabic characters have four forms :
413 * Isolated, Initial, Middle and Final Form
417 if(ch >= 0x0622 && ch <= 0x06D3) {
418 return(araLink[ch-0x0622]);
419 } else if(ch == 0x200D) {
421 } else if(ch >= 0x206D && ch <= 0x206F) {
423 }else if(ch >= 0xFB50 && ch <= 0xFC62) {
424 return(presALink[ch-0xFB50]);
425 } else if(ch >= 0xFE70 && ch <= 0xFEFC) {
426 return(presBLink[ch-0xFE70]);
434 *Function : Counts the number of spaces
435 * at each end of the logical buffer
438 countSpaces(UChar *dest,int32_t size,uint32_t options,int32_t *spacesCountl,int32_t *spacesCountr) {
440 int32_t countl = 0,countr = 0;
441 while(dest[i] == SPACE_CHAR) {
445 while(dest[size-1] == SPACE_CHAR) {
449 *spacesCountl = countl;
450 *spacesCountr = countr;
454 *Name : isTashkeelChar
455 *Function : Returns 1 for Tashkeel characters in 06 range else return 0
457 static U_INLINE int32_t
458 isTashkeelChar(UChar ch) {
459 return (int32_t)( ch>=0x064B && ch<= 0x0652 );
463 *Name : isTashkeelCharFE
464 *Function : Returns 1 for Tashkeel characters in FE range else return 0
466 static U_INLINE int32_t
467 isTashkeelCharFE(UChar ch) {
468 return (int32_t)( ch>=0xFE70 && ch<= 0xFE7F );
473 *Function : Returns 1 for Alef characters else return 0
475 static U_INLINE int32_t
476 isAlefChar(UChar ch) {
477 return (int32_t)( (ch==0x0622)||(ch==0x0623)||(ch==0x0625)||(ch==0x0627) );
481 *Name : isLamAlefChar
482 *Function : Returns 1 for LamAlef characters else return 0
484 static U_INLINE int32_t
485 isLamAlefChar(UChar ch) {
486 return (int32_t)((ch>=0xFEF5)&&(ch<=0xFEFC) );
491 *Function : returns 1 if the character matches one of the tail characters (0xfe73 or 0x200b) otherwise returns 0
494 static U_INLINE int32_t
495 isTailChar(UChar ch) {
496 if(ch == OLD_TAIL_CHAR || ch == NEW_TAIL_CHAR){
504 *Name : isSeenTailFamilyChar
505 *Function : returns 1 if the character is a seen family isolated character
506 * in the FE range otherwise returns 0
509 static U_INLINE int32_t
510 isSeenTailFamilyChar(UChar ch) {
511 if(ch >= 0xfeb1 && ch < 0xfebf){
512 return tailFamilyIsolatedFinal [ch - 0xFEB1];
518 /* Name : isSeenFamilyChar
519 * Function : returns 1 if the character is a seen family character in the Unicode
520 * 06 range otherwise returns 0
523 static U_INLINE int32_t
524 isSeenFamilyChar(UChar ch){
525 if(ch >= 0x633 && ch <= 0x636){
534 *Name : isAlefMaksouraChar
535 *Function : returns 1 if the character is a Alef Maksoura Final or isolated
536 * otherwise returns 0
538 static U_INLINE int32_t
539 isAlefMaksouraChar(UChar ch) {
540 return (int32_t)( (ch == 0xFEEF) || ( ch == 0xFEF0) || (ch == 0x0649));
544 * Name : isYehHamzaChar
545 * Function : returns 1 if the character is a yehHamza isolated or yehhamza
546 * final is found otherwise returns 0
548 static U_INLINE int32_t
549 isYehHamzaChar(UChar ch) {
550 if((ch==0xFE89)||(ch==0xFE8A)){
558 * Name: isTashkeelOnTatweelChar
559 * Function: Checks if the Tashkeel Character is on Tatweel or not,if the
560 * Tashkeel on tatweel (FE range), it returns 1 else if the
561 * Tashkeel with shadda on tatweel (FC range)return 2 otherwise
564 static U_INLINE int32_t
565 isTashkeelOnTatweelChar(UChar ch){
566 if(ch >= 0xfe70 && ch <= 0xfe7f && ch != NEW_TAIL_CHAR && ch != 0xFE75 && ch != SHADDA_TATWEEL_CHAR)
568 return tashkeelMedial [ch - 0xFE70];
569 }else if( (ch >= 0xfcf2 && ch <= 0xfcf4) || (ch == SHADDA_TATWEEL_CHAR)) {
577 * Name: isIsolatedTashkeelChar
578 * Function: Checks if the Tashkeel Character is in the isolated form
579 * (i.e. Unicode FE range) returns 1 else if the Tashkeel
580 * with shadda is in the isolated form (i.e. Unicode FC range)
581 * returns 2 otherwise returns 0
583 static U_INLINE int32_t
584 isIsolatedTashkeelChar(UChar ch){
585 if(ch >= 0xfe70 && ch <= 0xfe7f && ch != NEW_TAIL_CHAR && ch != 0xFE75){
586 return (1 - tashkeelMedial [ch - 0xFE70]);
587 }else if(ch >= 0xfc5e && ch <= 0xfc63){
598 *Name : calculateSize
599 *Function : This function calculates the destSize to be used in preflighting
600 * when the destSize is equal to 0
601 * It is used also to calculate the new destsize in case the
602 * destination buffer will be resized.
606 calculateSize(const UChar *source, int32_t sourceLength,
607 int32_t destSize,uint32_t options) {
610 int lamAlefOption = 0;
611 int tashkeelOption = 0;
613 destSize = sourceLength;
615 if (((options&U_SHAPE_LETTERS_MASK) == U_SHAPE_LETTERS_SHAPE ||
616 ((options&U_SHAPE_LETTERS_MASK) == U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED )) &&
617 ((options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_RESIZE )){
620 if((options&U_SHAPE_LETTERS_MASK) == U_SHAPE_LETTERS_SHAPE &&
621 ((options&U_SHAPE_TASHKEEL_MASK) == U_SHAPE_TASHKEEL_RESIZE ) ){
625 if(lamAlefOption || tashkeelOption){
626 if((options&U_SHAPE_TEXT_DIRECTION_MASK)==U_SHAPE_TEXT_DIRECTION_VISUAL_LTR) {
627 for(i=0;i<sourceLength;i++) {
628 if( ((isAlefChar(source[i]))&& (i<(sourceLength-1)) &&(source[i+1] == LAM_CHAR)) || (isTashkeelCharFE(source[i])) ) {
632 }else if((options&U_SHAPE_TEXT_DIRECTION_MASK)==U_SHAPE_TEXT_DIRECTION_LOGICAL) {
633 for(i=0;i<sourceLength;i++) {
634 if( ( (source[i] == LAM_CHAR) && (i<(sourceLength-1)) && (isAlefChar(source[i+1]))) || (isTashkeelCharFE(source[i])) ) {
641 if ((options&U_SHAPE_LETTERS_MASK) == U_SHAPE_LETTERS_UNSHAPE){
642 if ( (options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_RESIZE){
643 for(i=0;i<sourceLength;i++) {
644 if(isLamAlefChar(source[i]))
654 *Name : handleTashkeelWithTatweel
655 *Function : Replaces Tashkeel as following:
656 * Case 1 :if the Tashkeel on tatweel, replace it with Tatweel.
657 * Case 2 :if the Tashkeel aggregated with Shadda on Tatweel, replace
658 * it with Shadda on Tatweel.
659 * Case 3: if the Tashkeel is isolated replace it with Space.
663 handleTashkeelWithTatweel(UChar *dest, int32_t sourceLength,
664 int32_t destSize,uint32_t options,
665 UErrorCode *pErrorCode) {
667 for(i = 0; i < sourceLength; i++){
668 if((isTashkeelOnTatweelChar(dest[i]) == 1)){
669 dest[i] = TATWEEL_CHAR;
670 }else if((isTashkeelOnTatweelChar(dest[i]) == 2)){
671 dest[i] = SHADDA_TATWEEL_CHAR;
672 }else if(isIsolatedTashkeelChar(dest[i]) && dest[i] != SHADDA_CHAR){
673 dest[i] = SPACE_CHAR;
682 *Name : handleGeneratedSpaces
683 *Function : The shapeUnicode function converts Lam + Alef into LamAlef + space,
684 * and Tashkeel to space.
685 * handleGeneratedSpaces function puts these generated spaces
686 * according to the options the user specifies. LamAlef and Tashkeel
687 * spaces can be replaced at begin, at end, at near or decrease the
690 * There is also Auto option for LamAlef and tashkeel, which will put
691 * the spaces at end of the buffer (or end of text if the user used
692 * the option U_SHAPE_SPACES_RELATIVE_TO_TEXT_BEGIN_END).
694 * If the text type was visual_LTR and the option
695 * U_SHAPE_SPACES_RELATIVE_TO_TEXT_BEGIN_END was selected the END
696 * option will place the space at the beginning of the buffer and
697 * BEGIN will place the space at the end of the buffer.
701 handleGeneratedSpaces(UChar *dest, int32_t sourceLength,
704 UErrorCode *pErrorCode ) {
706 int32_t i = 0, j = 0;
708 UChar *tempbuffer=NULL;
710 int lamAlefOption = 0;
711 int tashkeelOption = 0;
712 int shapingMode = SHAPE_MODE;
714 if (shapingMode == 0){
715 if ( (options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_RESIZE ){
718 if ( (options&U_SHAPE_TASHKEEL_MASK) == U_SHAPE_TASHKEEL_RESIZE ){
723 if (lamAlefOption || tashkeelOption){
724 tempbuffer = (UChar *)uprv_malloc((sourceLength+1)*U_SIZEOF_UCHAR);
726 if(tempbuffer == NULL) {
727 *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
731 uprv_memset(tempbuffer, 0, (sourceLength+1)*U_SIZEOF_UCHAR);
733 i = j = 0; count = 0;
734 while(i < sourceLength) {
735 if ( (lamAlefOption && dest[i] == LAMALEF_SPACE_SUB) ||
736 (tashkeelOption && dest[i] == TASHKEEL_SPACE_SUB) ){
740 tempbuffer[j] = dest[i];
747 tempbuffer[i] = 0x0000;
752 uprv_memcpy(dest, tempbuffer, sourceLength*U_SIZEOF_UCHAR);
753 destSize = u_strlen(dest);
758 if (shapingMode == 0){
759 if ( (options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_NEAR ){
765 /* Lam+Alef is already shaped into LamAlef + FFFF */
767 while(i < sourceLength) {
768 if(lamAlefOption&&dest[i] == LAMALEF_SPACE_SUB){
769 dest[i] = SPACE_CHAR;
773 destSize = sourceLength;
778 if (shapingMode == 0) {
779 if ( ((options&U_SHAPE_LAMALEF_MASK) == uShapeLamalefBegin) ||
780 (((options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_AUTO )
781 && (spacesRelativeToTextBeginEnd==1)) ) {
784 if ( (options&U_SHAPE_TASHKEEL_MASK) == uShapeTashkeelBegin ) {
789 if(lamAlefOption || tashkeelOption){
790 tempbuffer = (UChar *)uprv_malloc((sourceLength+1)*U_SIZEOF_UCHAR);
793 if(tempbuffer == NULL) {
794 *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
798 uprv_memset(tempbuffer, 0, (sourceLength+1)*U_SIZEOF_UCHAR);
799 i = j = sourceLength; count = 0;
801 if ( (lamAlefOption && dest[i] == LAMALEF_SPACE_SUB) ||
802 (tashkeelOption && dest[i] == TASHKEEL_SPACE_SUB) ){
806 tempbuffer[j] = dest[i];
812 for(i=0 ;i < count; i++){
813 tempbuffer[i] = SPACE_CHAR;
816 uprv_memcpy(dest, tempbuffer, sourceLength*U_SIZEOF_UCHAR);
817 destSize = sourceLength;
825 if (shapingMode == 0) {
826 if ( ((options&U_SHAPE_LAMALEF_MASK) == uShapeLamalefEnd) ||
827 (((options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_AUTO )
828 && (spacesRelativeToTextBeginEnd==0)) ) {
831 if ( (options&U_SHAPE_TASHKEEL_MASK) == uShapeTashkeelEnd ){
836 if(lamAlefOption || tashkeelOption){
837 tempbuffer = (UChar *)uprv_malloc((sourceLength+1)*U_SIZEOF_UCHAR);
839 if(tempbuffer == NULL) {
840 *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
844 uprv_memset(tempbuffer, 0, (sourceLength+1)*U_SIZEOF_UCHAR);
846 i = j = 0; count = 0;
847 while(i < sourceLength) {
848 if ( (lamAlefOption && dest[i] == LAMALEF_SPACE_SUB) ||
849 (tashkeelOption && dest[i] == TASHKEEL_SPACE_SUB) ){
853 tempbuffer[j] = dest[i];
860 tempbuffer[i] = SPACE_CHAR;
865 uprv_memcpy(dest,tempbuffer, sourceLength*U_SIZEOF_UCHAR);
866 destSize = sourceLength;
871 uprv_free(tempbuffer);
878 *Name :expandCompositCharAtBegin
879 *Function :Expands the LamAlef character to Lam and Alef consuming the required
880 * space from beginning of the buffer. If the text type was visual_LTR
881 * and the option U_SHAPE_SPACES_RELATIVE_TO_TEXT_BEGIN_END was selected
882 * the spaces will be located at end of buffer.
883 * If there are no spaces to expand the LamAlef, an error
884 * will be set to U_NO_SPACE_AVAILABLE as defined in utypes.h
888 expandCompositCharAtBegin(UChar *dest, int32_t sourceLength, int32_t destSize,UErrorCode *pErrorCode) {
891 UChar *tempbuffer=NULL;
893 tempbuffer = (UChar *)uprv_malloc((sourceLength+1)*U_SIZEOF_UCHAR);
896 if(tempbuffer == NULL) {
897 *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
901 uprv_memset(tempbuffer, 0, (sourceLength+1)*U_SIZEOF_UCHAR);
904 while(dest[i] == SPACE_CHAR) {
909 i = j = sourceLength-1;
911 while(i >= 0 && j >= 0) {
912 if( countl>0 && isLamAlefChar(dest[i])) {
913 tempbuffer[j] = LAM_CHAR;
914 tempbuffer[j-1] = convertLamAlef[ dest[i] - 0xFEF5 ];
918 if( countl == 0 && isLamAlefChar(dest[i]) ) {
919 *pErrorCode=U_NO_SPACE_AVAILABLE;
921 tempbuffer[j] = dest[i];
926 uprv_memcpy(dest, tempbuffer, sourceLength*U_SIZEOF_UCHAR);
928 uprv_free(tempbuffer);
930 destSize = sourceLength;
935 *Name : expandCompositCharAtEnd
936 *Function : Expands the LamAlef character to Lam and Alef consuming the
937 * required space from end of the buffer. If the text type was
938 * Visual LTR and the option U_SHAPE_SPACES_RELATIVE_TO_TEXT_BEGIN_END
939 * was used, the spaces will be consumed from begin of buffer. If
940 * there are no spaces to expand the LamAlef, an error
941 * will be set to U_NO_SPACE_AVAILABLE as defined in utypes.h
945 expandCompositCharAtEnd(UChar *dest, int32_t sourceLength, int32_t destSize,UErrorCode *pErrorCode) {
949 int32_t inpsize = sourceLength;
951 UChar *tempbuffer=NULL;
952 tempbuffer = (UChar *)uprv_malloc((sourceLength+1)*U_SIZEOF_UCHAR);
955 if(tempbuffer == NULL) {
956 *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
960 uprv_memset(tempbuffer, 0, (sourceLength+1)*U_SIZEOF_UCHAR);
962 while(dest[inpsize-1] == SPACE_CHAR) {
967 i = sourceLength - countr - 1;
968 j = sourceLength - 1;
970 while(i >= 0 && j >= 0) {
971 if( countr>0 && isLamAlefChar(dest[i]) ) {
972 tempbuffer[j] = LAM_CHAR;
973 tempbuffer[j-1] = convertLamAlef[ dest[i] - 0xFEF5 ];
977 if ((countr == 0) && isLamAlefChar(dest[i]) ) {
978 *pErrorCode=U_NO_SPACE_AVAILABLE;
980 tempbuffer[j] = dest[i];
987 uprv_memmove(tempbuffer, tempbuffer+countr, sourceLength*U_SIZEOF_UCHAR);
988 if(u_strlen(tempbuffer) < sourceLength) {
989 for(i=sourceLength-1;i>=sourceLength-countr;i--) {
990 tempbuffer[i] = SPACE_CHAR;
994 uprv_memcpy(dest, tempbuffer, sourceLength*U_SIZEOF_UCHAR);
996 uprv_free(tempbuffer);
998 destSize = sourceLength;
1003 *Name : expandCompositCharAtNear
1004 *Function : Expands the LamAlef character into Lam + Alef, YehHamza character
1005 * into Yeh + Hamza, SeenFamily character into SeenFamily character
1006 * + Tail, while consuming the space next to the character.
1007 * If there are no spaces next to the character, an error
1008 * will be set to U_NO_SPACE_AVAILABLE as defined in utypes.h
1012 expandCompositCharAtNear(UChar *dest, int32_t sourceLength, int32_t destSize,UErrorCode *pErrorCode,
1013 int yehHamzaOption, int seenTailOption, int lamAlefOption) {
1017 UChar lamalefChar, yehhamzaChar;
1019 for(i = 0 ;i<=sourceLength-1;i++) {
1020 if (seenTailOption && isSeenTailFamilyChar(dest[i])) {
1021 if ((i>0) && (dest[i-1] == SPACE_CHAR) ) {
1022 dest[i-1] = tailChar;
1024 *pErrorCode=U_NO_SPACE_AVAILABLE;
1026 }else if(yehHamzaOption && (isYehHamzaChar(dest[i])) ) {
1027 if ((i>0) && (dest[i-1] == SPACE_CHAR) ) {
1028 yehhamzaChar = dest[i];
1029 dest[i] = yehHamzaToYeh[yehhamzaChar - YEH_HAMZAFE_CHAR];
1030 dest[i-1] = HAMZAFE_CHAR;
1033 *pErrorCode=U_NO_SPACE_AVAILABLE;
1035 }else if(lamAlefOption && isLamAlefChar(dest[i+1])) {
1036 if(dest[i] == SPACE_CHAR){
1037 lamalefChar = dest[i+1];
1038 dest[i+1] = LAM_CHAR;
1039 dest[i] = convertLamAlef[ lamalefChar - 0xFEF5 ];
1041 *pErrorCode=U_NO_SPACE_AVAILABLE;
1045 destSize = sourceLength;
1049 * Name : expandCompositChar
1050 * Function : LamAlef, need special handling, since it expands from one
1051 * character into two characters while shaping or deshaping.
1052 * In order to expand it, near or far spaces according to the
1053 * options user specifies. Also buffer size can be increased.
1055 * For SeenFamily characters and YehHamza only the near option is
1056 * supported, while for LamAlef we can take spaces from begin, end,
1057 * near or even increase the buffer size.
1058 * There is also the Auto option for LamAlef only, which will first
1059 * search for a space at end, begin then near, respectively.
1060 * If there are no spaces to expand these characters, an error will be set to
1061 * U_NO_SPACE_AVAILABLE as defined in utypes.h
1065 expandCompositChar(UChar *dest, int32_t sourceLength,
1066 int32_t destSize,uint32_t options,
1067 UErrorCode *pErrorCode, int shapingMode) {
1069 int32_t i = 0,j = 0;
1071 UChar *tempbuffer=NULL;
1072 int yehHamzaOption = 0;
1073 int seenTailOption = 0;
1074 int lamAlefOption = 0;
1076 if (shapingMode == 1){
1077 if ( (options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_AUTO){
1079 if(spacesRelativeToTextBeginEnd == 0) {
1080 destSize = expandCompositCharAtEnd(dest, sourceLength, destSize, pErrorCode);
1082 if(*pErrorCode == U_NO_SPACE_AVAILABLE) {
1083 *pErrorCode = U_ZERO_ERROR;
1084 destSize = expandCompositCharAtBegin(dest, sourceLength, destSize, pErrorCode);
1087 destSize = expandCompositCharAtBegin(dest, sourceLength, destSize, pErrorCode);
1089 if(*pErrorCode == U_NO_SPACE_AVAILABLE) {
1090 *pErrorCode = U_ZERO_ERROR;
1091 destSize = expandCompositCharAtEnd(dest, sourceLength, destSize, pErrorCode);
1095 if(*pErrorCode == U_NO_SPACE_AVAILABLE) {
1096 *pErrorCode = U_ZERO_ERROR;
1097 destSize = expandCompositCharAtNear(dest, sourceLength, destSize, pErrorCode, yehHamzaOption,
1103 if (shapingMode == 1){
1104 if ( (options&U_SHAPE_LAMALEF_MASK) == uShapeLamalefEnd){
1105 destSize = expandCompositCharAtEnd(dest, sourceLength, destSize, pErrorCode);
1109 if (shapingMode == 1){
1110 if ( (options&U_SHAPE_LAMALEF_MASK) == uShapeLamalefBegin){
1111 destSize = expandCompositCharAtBegin(dest, sourceLength, destSize, pErrorCode);
1115 if (shapingMode == 0){
1116 if ((options&U_SHAPE_YEHHAMZA_MASK) == U_SHAPE_YEHHAMZA_TWOCELL_NEAR){
1119 if ((options&U_SHAPE_SEEN_MASK) == U_SHAPE_SEEN_TWOCELL_NEAR){
1123 if (shapingMode == 1) {
1124 if ( (options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_NEAR) {
1130 if (yehHamzaOption || seenTailOption || lamAlefOption){
1131 destSize = expandCompositCharAtNear(dest, sourceLength, destSize, pErrorCode, yehHamzaOption,
1132 seenTailOption,lamAlefOption);
1136 if (shapingMode == 1){
1137 if ( (options&U_SHAPE_LAMALEF_MASK) == U_SHAPE_LAMALEF_RESIZE){
1138 destSize = calculateSize(dest,sourceLength,destSize,options);
1139 tempbuffer = (UChar *)uprv_malloc((destSize+1)*U_SIZEOF_UCHAR);
1142 if(tempbuffer == NULL) {
1143 *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
1147 uprv_memset(tempbuffer, 0, (destSize+1)*U_SIZEOF_UCHAR);
1150 while(i < destSize && j < destSize) {
1151 if(isLamAlefChar(dest[i]) ) {
1152 tempbuffer[j] = convertLamAlef[ dest[i] - 0xFEF5 ];
1153 tempbuffer[j+1] = LAM_CHAR;
1156 tempbuffer[j] = dest[i];
1162 uprv_memcpy(dest, tempbuffer, destSize*U_SIZEOF_UCHAR);
1167 uprv_free(tempbuffer);
1173 *Name : shapeUnicode
1174 *Function : Converts an Arabic Unicode buffer in 06xx Range into a shaped
1175 * arabic Unicode buffer in FExx Range
1178 shapeUnicode(UChar *dest, int32_t sourceLength,
1179 int32_t destSize,uint32_t options,
1180 UErrorCode *pErrorCode,
1185 int32_t lastPos,Nx, Nw;
1187 int32_t lamalef_found = 0;
1188 int32_t seenfamFound = 0, yehhamzaFound =0, tashkeelFound = 0;
1189 UChar prevLink = 0, lastLink = 0, currLink, nextLink = 0;
1193 * Converts the input buffer from FExx Range into 06xx Range
1194 * to make sure that all characters are in the 06xx range
1195 * even the lamalef is converted to the special region in
1198 if ((options & U_SHAPE_PRESERVE_PRESENTATION_MASK) == U_SHAPE_PRESERVE_PRESENTATION_NOOP) {
1199 for (i = 0; i < sourceLength; i++) {
1200 UChar inputChar = dest[i];
1201 if ( (inputChar >= 0xFB50) && (inputChar <= 0xFBFF)) {
1202 UChar c = convertFBto06 [ (inputChar - 0xFB50) ];
1205 } else if ( (inputChar >= 0xFE70) && (inputChar <= 0xFEFC)) {
1206 dest[i] = convertFEto06 [ (inputChar - 0xFE70) ] ;
1208 dest[i] = inputChar ;
1214 /* sets the index to the end of the buffer, together with the step point to -1 */
1215 i = sourceLength - 1;
1220 * This function resolves the link between the characters .
1221 * Arabic characters have four forms :
1222 * Isolated Form, Initial Form, Middle Form and Final Form
1224 currLink = getLink(dest[i]);
1230 /* If high byte of currLink > 0 then more than one shape */
1231 if ((currLink & 0xFF00) > 0 || (getLink(dest[i]) & IRRELEVANT) != 0) {
1233 while (Nx < 0) { /* we need to know about next char */
1238 nextLink = getLink(dest[Nw]);
1239 if((nextLink & IRRELEVANT) == 0) {
1247 if ( ((currLink & ALEFTYPE) > 0) && ((lastLink & LAMTYPE) > 0) ) {
1249 wLamalef = changeLamAlef(dest[i]); /*get from 0x065C-0x065f */
1250 if ( wLamalef != 0) {
1251 dest[i] = LAMALEF_SPACE_SUB; /* The default case is to drop the Alef and replace */
1252 dest[lastPos] =wLamalef; /* it by LAMALEF_SPACE_SUB which is the last character in the */
1253 i=lastPos; /* unicode private use area, this is done to make */
1254 } /* sure that removeLamAlefSpaces() handles only the */
1255 lastLink = prevLink; /* spaces generated during lamalef generation. */
1256 currLink = getLink(wLamalef); /* LAMALEF_SPACE_SUB is added here and is replaced by spaces */
1257 } /* in removeLamAlefSpaces() */
1259 if ((i > 0) && (dest[i-1] == SPACE_CHAR)){
1260 if ( isSeenFamilyChar(dest[i])){
1262 } else if (dest[i] == YEH_HAMZA_CHAR) {
1267 if ( isSeenFamilyChar(dest[i])){
1269 } else if (dest[i] == YEH_HAMZA_CHAR) {
1275 * get the proper shape according to link ability of neighbors
1276 * and of character; depends on the order of the shapes
1277 * (isolated, initial, middle, final) in the compatibility area
1279 Shape = shapeTable[nextLink & (LINKR + LINKL)]
1280 [lastLink & (LINKR + LINKL)]
1281 [currLink & (LINKR + LINKL)];
1283 if ((currLink & (LINKR+LINKL)) == 1) {
1285 } else if(isTashkeelChar(dest[i])) {
1286 if( (lastLink & LINKL) && (nextLink & LINKR) && (tashkeelFlag == 1) &&
1287 dest[i] != 0x064C && dest[i] != 0x064D )
1290 if( (nextLink&ALEFTYPE) == ALEFTYPE && (lastLink&LAMTYPE) == LAMTYPE ) {
1298 if ((dest[i] ^ 0x0600) < 0x100) {
1299 if ( isTashkeelChar(dest[i]) ){
1300 if (tashkeelFlag == 2){
1301 dest[i] = TASHKEEL_SPACE_SUB;
1304 dest[i] = 0xFE70 + IrrelevantPos[(dest[i] - 0x064B)] + Shape;
1306 }else if ((currLink & APRESENT) > 0) {
1307 dest[i] = (UChar)(0xFB50 + (currLink >> 8) + Shape);
1308 }else if ((currLink >> 8) > 0 && (currLink & IRRELEVANT) == 0) {
1309 dest[i] = (UChar)(0xFE70 + (currLink >> 8) + Shape);
1314 /* move one notch forward */
1315 if ((currLink & IRRELEVANT) == 0) {
1316 prevLink = lastLink;
1317 lastLink = currLink;
1323 currLink = nextLink;
1325 } else if(i != iend) {
1326 currLink = getLink(dest[i]);
1329 destSize = sourceLength;
1330 if ( (lamalef_found != 0 ) || (tashkeelFound != 0) ){
1331 destSize = handleGeneratedSpaces(dest,sourceLength,destSize,options,pErrorCode);
1334 if ( (seenfamFound != 0) || (yehhamzaFound != 0) ) {
1335 destSize = expandCompositChar(dest, sourceLength,destSize,options,pErrorCode, SHAPE_MODE);
1341 *Name : deShapeUnicode
1342 *Function : Converts an Arabic Unicode buffer in FExx Range into unshaped
1343 * arabic Unicode buffer in 06xx Range
1346 deShapeUnicode(UChar *dest, int32_t sourceLength,
1347 int32_t destSize,uint32_t options,
1348 UErrorCode *pErrorCode) {
1350 int32_t lamalef_found = 0;
1351 int32_t yehHamzaComposeEnabled = 0;
1352 int32_t seenComposeEnabled = 0;
1354 yehHamzaComposeEnabled = ((options&U_SHAPE_YEHHAMZA_MASK) == U_SHAPE_YEHHAMZA_TWOCELL_NEAR) ? 1 : 0;
1355 seenComposeEnabled = ((options&U_SHAPE_SEEN_MASK) == U_SHAPE_SEEN_TWOCELL_NEAR)? 1 : 0;
1358 *This for loop changes the buffer from the Unicode FE range to
1359 *the Unicode 06 range
1362 for(i = 0; i < sourceLength; i++) {
1363 UChar inputChar = dest[i];
1364 if ( (inputChar >= 0xFB50) && (inputChar <= 0xFBFF)) { /* FBxx Arabic range */
1365 UChar c = convertFBto06 [ (inputChar - 0xFB50) ];
1368 } else if( (yehHamzaComposeEnabled == 1) && ((inputChar == HAMZA06_CHAR) || (inputChar == HAMZAFE_CHAR))
1369 && (i < (sourceLength - 1)) && isAlefMaksouraChar(dest[i+1] )) {
1370 dest[i] = SPACE_CHAR;
1371 dest[i+1] = YEH_HAMZA_CHAR;
1372 } else if ( (seenComposeEnabled == 1) && (isTailChar(inputChar)) && (i< (sourceLength - 1))
1373 && (isSeenTailFamilyChar(dest[i+1])) ) {
1374 dest[i] = SPACE_CHAR;
1375 } else if (( inputChar >= 0xFE70) && (inputChar <= 0xFEF4 )) { /* FExx Arabic range */
1376 dest[i] = convertFEto06 [ (inputChar - 0xFE70) ];
1378 dest[i] = inputChar ;
1381 if( isLamAlefChar(dest[i]) )
1385 destSize = sourceLength;
1386 if (lamalef_found != 0){
1387 destSize = expandCompositChar(dest,sourceLength,destSize,options,pErrorCode,DESHAPE_MODE);
1393 ****************************************
1395 ****************************************
1398 U_CAPI int32_t U_EXPORT2
1399 u_shapeArabic(const UChar *source, int32_t sourceLength,
1400 UChar *dest, int32_t destCapacity,
1402 UErrorCode *pErrorCode) {
1406 spacesRelativeToTextBeginEnd = 0;
1407 uShapeLamalefBegin = U_SHAPE_LAMALEF_BEGIN;
1408 uShapeLamalefEnd = U_SHAPE_LAMALEF_END;
1409 uShapeTashkeelBegin = U_SHAPE_TASHKEEL_BEGIN;
1410 uShapeTashkeelEnd = U_SHAPE_TASHKEEL_END;
1412 /* usual error checking */
1413 if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
1417 /* make sure that no reserved options values are used; allow dest==NULL only for preflighting */
1418 if( source==NULL || sourceLength<-1 || (dest==NULL && destCapacity!=0) || destCapacity<0 ||
1419 (((options&U_SHAPE_TASHKEEL_MASK) > 0) &&
1420 ((options&U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED) == U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED) ) ||
1421 (((options&U_SHAPE_TASHKEEL_MASK) > 0) &&
1422 ((options&U_SHAPE_LETTERS_MASK) == U_SHAPE_LETTERS_UNSHAPE)) ||
1423 (options&U_SHAPE_DIGIT_TYPE_RESERVED)==U_SHAPE_DIGIT_TYPE_RESERVED ||
1424 (options&U_SHAPE_DIGITS_MASK)==U_SHAPE_DIGITS_RESERVED ||
1425 ((options&U_SHAPE_LAMALEF_MASK) != U_SHAPE_LAMALEF_RESIZE &&
1426 (options&U_SHAPE_AGGREGATE_TASHKEEL_MASK) != 0) ||
1427 ((options&U_SHAPE_AGGREGATE_TASHKEEL_MASK) == U_SHAPE_AGGREGATE_TASHKEEL &&
1428 (options&U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED) != U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED)
1431 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1434 /* Validate lamalef options */
1435 if(((options&U_SHAPE_LAMALEF_MASK) > 0)&&
1436 !(((options & U_SHAPE_LAMALEF_MASK)==U_SHAPE_LAMALEF_BEGIN) ||
1437 ((options & U_SHAPE_LAMALEF_MASK)==U_SHAPE_LAMALEF_END ) ||
1438 ((options & U_SHAPE_LAMALEF_MASK)==U_SHAPE_LAMALEF_RESIZE )||
1439 ((options & U_SHAPE_LAMALEF_MASK)==U_SHAPE_LAMALEF_AUTO) ||
1440 ((options & U_SHAPE_LAMALEF_MASK)==U_SHAPE_LAMALEF_NEAR)))
1442 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1445 /* Validate Tashkeel options */
1446 if(((options&U_SHAPE_TASHKEEL_MASK) > 0)&&
1447 !(((options & U_SHAPE_TASHKEEL_MASK)==U_SHAPE_TASHKEEL_BEGIN) ||
1448 ((options & U_SHAPE_TASHKEEL_MASK)==U_SHAPE_TASHKEEL_END )
1449 ||((options & U_SHAPE_TASHKEEL_MASK)==U_SHAPE_TASHKEEL_RESIZE )||
1450 ((options & U_SHAPE_TASHKEEL_MASK)==U_SHAPE_TASHKEEL_REPLACE_BY_TATWEEL)))
1452 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1455 /* determine the source length */
1456 if(sourceLength==-1) {
1457 sourceLength=u_strlen(source);
1459 if(sourceLength<=0) {
1460 return u_terminateUChars(dest, destCapacity, 0, pErrorCode);
1463 /* check that source and destination do not overlap */
1465 ((source<=dest && dest<source+sourceLength) ||
1466 (dest<=source && source<dest+destCapacity))) {
1467 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1471 /* Does Options contain the new Seen Tail Unicode code point option */
1472 if ( (options&SHAPE_TAIL_TYPE_MASK) == SHAPE_TAIL_NEW_UNICODE){
1473 tailChar = NEW_TAIL_CHAR;
1475 tailChar = OLD_TAIL_CHAR;
1478 if((options&U_SHAPE_LETTERS_MASK)!=U_SHAPE_LETTERS_NOOP) {
1480 UChar *tempbuffer, *tempsource = NULL;
1481 int32_t outputSize, spacesCountl=0, spacesCountr=0;
1483 if((options&U_SHAPE_AGGREGATE_TASHKEEL_MASK)>0) {
1484 int32_t logical_order = (options&U_SHAPE_TEXT_DIRECTION_MASK) == U_SHAPE_TEXT_DIRECTION_LOGICAL;
1485 int32_t aggregate_tashkeel =
1486 (options&(U_SHAPE_AGGREGATE_TASHKEEL_MASK+U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED)) ==
1487 (U_SHAPE_AGGREGATE_TASHKEEL+U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED);
1488 int step=logical_order?1:-1;
1489 int j=logical_order?-1:2*sourceLength;
1490 int i=logical_order?-1:sourceLength;
1491 int end=logical_order?sourceLength:-1;
1492 int aggregation_possible = 1;
1494 UChar prevLink, currLink = 0;
1495 int newSourceLength = 0;
1496 tempsource = (UChar *)uprv_malloc(2*sourceLength*U_SIZEOF_UCHAR);
1497 if(tempsource == NULL) {
1498 *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
1502 while ((i+=step) != end) {
1503 prevLink = currLink;
1504 currLink = getLink(source[i]);
1505 if (aggregate_tashkeel && ((prevLink|currLink)&COMBINE) == COMBINE && aggregation_possible) {
1506 aggregation_possible = 0;
1507 tempsource[j] = (prev<source[i]?prev:source[i])-0x064C+0xFC5E;
1508 currLink = getLink(tempsource[j]);
1510 aggregation_possible = 1;
1511 tempsource[j+=step] = source[i];
1516 source = tempsource+(logical_order?0:j);
1517 sourceLength = newSourceLength;
1520 /* calculate destination size */
1521 /* TODO: do we ever need to do this pure preflighting? */
1522 if(((options&U_SHAPE_LAMALEF_MASK)==U_SHAPE_LAMALEF_RESIZE) ||
1523 ((options&U_SHAPE_TASHKEEL_MASK)==U_SHAPE_TASHKEEL_RESIZE)) {
1524 outputSize=calculateSize(source,sourceLength,destCapacity,options);
1526 outputSize=sourceLength;
1529 if(outputSize>destCapacity) {
1530 *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
1531 if (tempsource != NULL) uprv_free(tempsource);
1536 * need a temporary buffer of size max(outputSize, sourceLength)
1537 * because at first we copy source->temp
1539 if(sourceLength>outputSize) {
1540 outputSize=sourceLength;
1543 /* Start of Arabic letter shaping part */
1544 if(outputSize<=sizeof(buffer)/U_SIZEOF_UCHAR) {
1545 outputSize=sizeof(buffer)/U_SIZEOF_UCHAR;
1548 tempbuffer = (UChar *)uprv_malloc(outputSize*U_SIZEOF_UCHAR);
1551 if(tempbuffer == NULL) {
1552 *pErrorCode = U_MEMORY_ALLOCATION_ERROR;
1553 if (tempsource != NULL) uprv_free(tempsource);
1557 uprv_memcpy(tempbuffer, source, sourceLength*U_SIZEOF_UCHAR);
1558 if (tempsource != NULL){
1559 uprv_free(tempsource);
1562 if(sourceLength<outputSize) {
1563 uprv_memset(tempbuffer+sourceLength, 0, (outputSize-sourceLength)*U_SIZEOF_UCHAR);
1566 if((options&U_SHAPE_TEXT_DIRECTION_MASK) == U_SHAPE_TEXT_DIRECTION_LOGICAL) {
1567 countSpaces(tempbuffer,sourceLength,options,&spacesCountl,&spacesCountr);
1568 invertBuffer(tempbuffer,sourceLength,options,spacesCountl,spacesCountr);
1571 if((options&U_SHAPE_TEXT_DIRECTION_MASK) == U_SHAPE_TEXT_DIRECTION_VISUAL_LTR) {
1572 if((options&U_SHAPE_SPACES_RELATIVE_TO_TEXT_MASK) == U_SHAPE_SPACES_RELATIVE_TO_TEXT_BEGIN_END) {
1573 spacesRelativeToTextBeginEnd = 1;
1574 uShapeLamalefBegin = U_SHAPE_LAMALEF_END;
1575 uShapeLamalefEnd = U_SHAPE_LAMALEF_BEGIN;
1577 uShapeTashkeelBegin = U_SHAPE_TASHKEEL_END;
1578 uShapeTashkeelEnd = U_SHAPE_TASHKEEL_BEGIN;
1582 switch(options&U_SHAPE_LETTERS_MASK) {
1583 case U_SHAPE_LETTERS_SHAPE :
1584 if( (options&U_SHAPE_TASHKEEL_MASK)> 0
1585 && ((options&U_SHAPE_TASHKEEL_MASK) !=U_SHAPE_TASHKEEL_REPLACE_BY_TATWEEL)) {
1586 /* Call the shaping function with tashkeel flag == 2 for removal of tashkeel */
1587 destLength = shapeUnicode(tempbuffer,sourceLength,destCapacity,options,pErrorCode,2);
1589 /* default Call the shaping function with tashkeel flag == 1 */
1590 destLength = shapeUnicode(tempbuffer,sourceLength,destCapacity,options,pErrorCode,1);
1592 /*After shaping text check if user wants to remove tashkeel and replace it with tatweel*/
1593 if( (options&U_SHAPE_TASHKEEL_MASK) == U_SHAPE_TASHKEEL_REPLACE_BY_TATWEEL){
1594 destLength = handleTashkeelWithTatweel(tempbuffer,destLength,destCapacity,options,pErrorCode);
1598 case U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED :
1599 /* Call the shaping function with tashkeel flag == 0 */
1600 destLength = shapeUnicode(tempbuffer,sourceLength,destCapacity,options,pErrorCode,0);
1603 case U_SHAPE_LETTERS_UNSHAPE :
1604 /* Call the deshaping function */
1605 destLength = deShapeUnicode(tempbuffer,sourceLength,destCapacity,options,pErrorCode);
1608 /* will never occur because of validity checks above */
1614 * TODO: (markus 2002aug01)
1615 * For as long as we always preflight the outputSize above
1616 * we should U_ASSERT(outputSize==destLength)
1617 * except for the adjustment above before the tempbuffer allocation
1620 if((options&U_SHAPE_TEXT_DIRECTION_MASK) == U_SHAPE_TEXT_DIRECTION_LOGICAL) {
1621 countSpaces(tempbuffer,destLength,options,&spacesCountl,&spacesCountr);
1622 invertBuffer(tempbuffer,destLength,options,spacesCountl,spacesCountr);
1624 uprv_memcpy(dest, tempbuffer, uprv_min(destLength, destCapacity)*U_SIZEOF_UCHAR);
1626 if(tempbuffer!=buffer) {
1627 uprv_free(tempbuffer);
1630 if(destLength>destCapacity) {
1631 *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
1635 /* End of Arabic letter shaping part */
1638 * No letter shaping:
1639 * just make sure the destination is large enough and copy the string.
1641 if(destCapacity<sourceLength) {
1642 /* this catches preflighting, too */
1643 *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
1644 return sourceLength;
1646 uprv_memcpy(dest, source, sourceLength*U_SIZEOF_UCHAR);
1647 destLength=sourceLength;
1651 * Perform number shaping.
1652 * With UTF-16 or UTF-32, the length of the string is constant.
1653 * The easiest way to do this is to operate on the destination and
1654 * "shape" the digits in-place.
1656 if((options&U_SHAPE_DIGITS_MASK)!=U_SHAPE_DIGITS_NOOP) {
1660 /* select the requested digit group */
1661 switch(options&U_SHAPE_DIGIT_TYPE_MASK) {
1662 case U_SHAPE_DIGIT_TYPE_AN:
1663 digitBase=0x660; /* Unicode: "Arabic-Indic digits" */
1665 case U_SHAPE_DIGIT_TYPE_AN_EXTENDED:
1666 digitBase=0x6f0; /* Unicode: "Eastern Arabic-Indic digits (Persian and Urdu)" */
1669 /* will never occur because of validity checks above */
1674 /* perform the requested operation */
1675 switch(options&U_SHAPE_DIGITS_MASK) {
1676 case U_SHAPE_DIGITS_EN2AN:
1677 /* add (digitBase-'0') to each European (ASCII) digit code point */
1679 for(i=0; i<destLength; ++i) {
1680 if(((uint32_t)dest[i]-0x30)<10) {
1685 case U_SHAPE_DIGITS_AN2EN:
1686 /* subtract (digitBase-'0') from each Arabic digit code point */
1687 for(i=0; i<destLength; ++i) {
1688 if(((uint32_t)dest[i]-(uint32_t)digitBase)<10) {
1689 dest[i]-=digitBase-0x30;
1693 case U_SHAPE_DIGITS_ALEN2AN_INIT_LR:
1694 _shapeToArabicDigitsWithContext(dest, destLength,
1696 (UBool)((options&U_SHAPE_TEXT_DIRECTION_MASK)==U_SHAPE_TEXT_DIRECTION_LOGICAL),
1699 case U_SHAPE_DIGITS_ALEN2AN_INIT_AL:
1700 _shapeToArabicDigitsWithContext(dest, destLength,
1702 (UBool)((options&U_SHAPE_TEXT_DIRECTION_MASK)==U_SHAPE_TEXT_DIRECTION_LOGICAL),
1706 /* will never occur because of validity checks above */
1711 return u_terminateUChars(dest, destCapacity, destLength, pErrorCode);